I think I might have something! I was running the chip at 10V, which was wrong. This chip runs on 15V (checked the output of the converter on the game's PCB... 14.8V). I bumped it up to 15V and now the test mode appears to be dumping the ROM! I have a whack of data coming out the RIOx pins, about 20.8ms long with a bit time of approximately 10.2us. This divides out to around 2040ish bits!
The chip has 2K of data so this is pretty close, well within the measurement error of my scope.
The voltage range for data dumping is fairly critical and I thought I saw something earlier while it was running at 10V. At around 12V on test, the chip dumps its data it looks. 11.6V to around 12.4V. Below 11.6V it just relays data from the input ports. (possibly throwing data into the bus to execute). Above 12.4V it runs the code in the ROM.
Going by PMOS standards, this would be -2.6V to -3.4V.
Looks like I will have to make yet another dumping rig for these- I will make it a three play. MM78, COP411, and the mitsubishi CPUs maybe. Will be good to get my allpro 88 going. I got a replacement PCB for it about half done.
Good news! Are you putting anything on any ports to get it to dump, or just setting the test pin and clocking the OSC pin?